Privacy Policy

Last updated: 5 October 2026 · Version 2026-10-05

Operated by Prospera Technologies FZ-LLC
Licensed by the Ras Al Khaimah Economic Zone (RAKEZ), United Arab Emirates
Contact: hello@tovira.io

1. Who we are and how to reach us

Tovira is operated by Prospera Technologies FZ-LLC ("Prospera", "we", "us"). Anything in this policy — questions, requests, complaints — goes to hello@tovira.io.

2. What you should know before you start

Tovira reads conversations you have had with your clients — voice notes you record and chat exports you upload — and turns them into structured facts you can search and be reminded of.

Three things follow from that, and we would rather say them plainly:

We hold messages written by your clients. They are not our users. They have not agreed to anything with us. Section 10 explains what that means and what we do about it.

You decide what goes in, and you are responsible for having the right to upload it. We act on your instructions. We do not select, review, or verify the conversations you give us. Section 4 of the Terms of Service sets out your responsibilities.

We do not keep your conversations to train AI. We keep anonymous statistics about how the product is used. Your conversations are kept so that the service can search and quote them for you — not to train models — and they are deleted when you close your account. Section 8 sets this out precisely.

If any of that is not acceptable to you, do not upload client conversations.

3. What we collect

When you create an account. Your name, email address, a password (stored only as a cryptographic hash, never in readable form), and a referral code if you enter one.

Payment information. Handled entirely by Stripe. We never see or store your card number. We keep your Stripe customer and subscription identifiers, your plan, and your billing period dates.

What you capture.

• Voice notes you record inside Tovira, and the text transcribed from them. Tovira provides no way to add an audio file from anywhere else: there is no upload, no share-in from another app, and any audio inside a chat export you upload is discarded without being read or transcribed. A voice note is limited to about ten minutes, because it is meant as your own note after a meeting.
• Chat exports you upload as .txt or .zip, including the full message history inside the file
• Text you paste or type
• Images you add from your device
• Client records you create, including names and contact details
• Inventory items you add, with descriptions and quantities

What we derive from it. Structured facts: requirements, commitments and their due dates, meetings, unanswered questions, the people mentioned, inventory matches, deal outcomes, and numerical representations of text used for search (embeddings). Every derived fact is stored alongside the exact message it came from, so you can always check it.

Your acknowledgements. When you first upload a chat export we ask you to confirm you have the right to do so. We record that confirmation and when you gave it.

Technical and usage information. Server logs, device and browser information, IP address, error reports, and counts of how much processing your account has used.

When you ask for access to Tovira. Access is currently by request rather than open signup. If you ask for access we collect your name, work email address, phone number, company name, your role, whether you own or are employed by that company, your trade licence number if you own it, what kind of client conversations you intend to upload, roughly how many, and the confirmation you tick before submitting. We also record your IP address and browser information, and the referral code if you arrived through one.

If you are employed rather than an owner, we ask you by email for your company's no-objection certificate. That document is held in the email correspondence and is used only to decide whether to grant you access.

We use all of this to decide whether to approve your request and to create your account if we do. We do not use it for marketing and we do not share it.

4. What we deliberately do not collect

We do not record health information as facts. Tovira is instructed never to record anything about a person's health, and any health fact it produces is automatically discarded before it is stored. If a message mentions a medical matter, that message stays in the conversation history it came from. On rare occasions a conversation summary may refer to it; we measure how often this happens and work to prevent it.

We remove high-risk numbers from text before it is stored. Every voice transcript, chat export and pasted message passes through automatic redaction before it is saved and before it is sent to the supplier that reads it and extracts facts. It removes Emirates ID numbers, card numbers, IBANs and bank account numbers, passport, visa and licence numbers, and login credentials, replacing them with a placeholder such as "[card ending 4421]". The original numbers are never stored in the text and never reach that supplier.

Three limits we would rather state than let you discover.

A recording has to be transcribed before it can be redacted. If you read a card number aloud in a voice note, the recording reaches our transcription supplier as you spoke it. The number is removed from the transcript before the transcript is stored, and it never reaches the supplier that extracts facts. But the recording itself was transmitted. This is inherent to every speech-to-text service and cannot be engineered away. Our transcription supplier is contracted not to train on what we send it.

We do not keep the recording. Once a voice note has been transcribed, the recording itself is deleted about thirty days later, and only the text remains. It may go sooner: an erasure request or closing your account removes it immediately. One exception, and we would rather name it than leave it implied — if a recording never transcribes successfully, we keep it, because the recording is then the only copy of what you captured and deleting it would destroy your note.

Images are stored exactly as you upload them. A photograph is not redacted, not scanned, and not read by any AI supplier. If you photograph a passport page or an Emirates ID card and add it to a client, that image is stored as it is and shown back to you. It stays in our storage, is visible only to you, and is never transmitted to any outside model. Please do not upload documents you would not want stored.

Redaction is thorough, not perfect. Some numbers are caught by their own structure — an Emirates ID, a valid card number, a UAE IBAN. Others are only caught when a nearby word identifies them, such as "account number" or "passport". A bare string of digits with nothing around it to identify it is left alone, because removing every unlabelled number would also remove prices, plot numbers and areas, and the service would stop working. We test this regularly against deliberately difficult examples, we record the result, and we publish nothing that implies it catches everything. Please do not treat redaction as a reason to send sensitive numbers through the product.

We do not take in recordings made elsewhere. Tovira has no feature for adding an audio file from another app, a file picker, or a share sheet. Voice capture exists so you can make your own note after a meeting. Using it to reproduce a recording of another person is prohibited by our Terms.

We do not track you across other websites. We do not sell data to anyone, for any purpose, ever.

5. Why we process your data and on what basis

Where UAE law requires consent as the default basis for processing, we rely on the exceptions available under that law for the activities above — in particular, processing necessary to perform our contract with you, processing necessary to comply with a legal obligation, and processing necessary for our legitimate interests where those do not override your rights. Where we rely on your consent, you may withdraw it at any time.

6. Where your data is held, and why it is outside the UAE

Your data is stored in Sweden, in Amazon Web Services' Stockholm region (eu-north-1), inside the European Union.

We host outside the UAE deliberately, for the availability and integrity of your data. During 2026 the AWS UAE region suffered repeated physical damage and extended outages, including incidents in March and July in which data centre facilities were struck and availability zones taken offline for prolonged periods. Our first obligation is never to lose a capture. We cannot meet that obligation from infrastructure that has been physically compromised twice in one year. Sweden was chosen because it sits inside the European Union's data protection regime, which is among the strictest in the world.

Where we transfer your data outside the UAE, we take steps to ensure it receives a level of protection appropriate to the sensitivity of the data, including choosing suppliers and locations, such as the European Union, that are subject to comprehensive data protection law, and entering into appropriate contractual protections with our suppliers.

7. How long we keep things

We considered deleting raw conversation content on a fixed schedule and decided against it. Searching your own history, asking questions about past conversations, and drafting follow-ups that reference what was actually said all depend on that content existing. A retention clock would silently break those features on older conversations without telling you why. Instead, we limit what we keep (no health facts, identification numbers removed before storage), we do not use it to train AI models, and we delete everything completely when you close your account.

8. What we learn from usage, and what we do not keep

What we keep long-term: anonymous numbers. We measure how the product is used so we can improve it and, in future, offer users evidence-based guidance on what works. The measurements are counts and timings — for example, how quickly replies happen, how many follow-ups occur before contact stops, how often commitments are kept, and whether a deal was eventually marked won or lost.

These are only produced above a minimum group size. We never compute or retain a statistic from a group of fewer than 20 reps or clients — small enough that it could otherwise identify an individual. Below that floor, no statistic is produced.

What we do not keep: your conversations, for training. We do not retain conversation content to train AI models. Your conversations are kept so that the service can search, quote and reason over them for you, and for no other purpose. They are deleted when you close your account.

What we do not do at all. We do not share your data, or statistics derived from it, with other Tovira users in any form that could identify you or your clients. We do not sell it. We do not use your clients' messages to build profiles of those clients for any purpose outside your own account.

9. Automated processing

Tovira uses automated language models to read your captures and produce structured facts, and to rank what to show you each day. Some of what it shows you is inferred rather than stated — for example, treating a long silence as a sign that a deal has gone cold. Inferences are labelled as such and you can correct them.

These systems never act on your behalf. Tovira does not send messages, contact clients, book meetings, share your inventory, or take any action in the outside world. It drafts and suggests. You decide.

Nothing Tovira produces is a decision with legal or similarly significant effects about you or anyone else.

10. If your messages are in Tovira but you are not a user

If someone you have exchanged messages with uses Tovira, your messages to them may be stored and processed by us on their behalf.

We did not get your messages from you. The Tovira user who uploaded them decided to do so. Under data protection law that user is responsible for having a lawful basis to share your messages with a service like ours, and they confirmed to us that they had one. We act on their instructions.

What we do to protect you regardless:

• We do not record health information about you as facts
• ID, card and bank numbers are removed from the text of your messages before they are stored, subject to the limits described in section 4
• Tovira has no feature for adding an audio recording from another app, and our Terms prohibit users from uploading a recording of another person. Users may only record their own notes inside the app.
• Photographs are not redacted. They are stored as uploaded and are never sent to any AI supplier.
• We never use your messages to train AI models
• We never contact you, market to you, or build a profile of you for any purpose outside that user's own account
• We never share your messages with anyone else

Your rights. You may contact us at hello@tovira.io to ask what we hold about you, to have it corrected, or to have it erased.

How we handle an erasure request. When you ask us to erase your data from a user's account, we tell that user a legal request has been received, and we restrict processing of the disputed data while we review it. They have 14 days to tell us of a legal basis for keeping it, identifying the specific data, the reason, how long it is needed, and why restricting or partly deleting it is not enough. Their objection is not a veto: we assess whether an exception genuinely applies. If we accept it, we tell you so, why, and how to complain to the regulator. Otherwise we proceed.

What we remove:

• Your own messages
• Facts recorded about you: your entry as a person, personal facts about you, and questions you asked
• Records of you in our processing logs and archives

What a person reviews. Some facts about you may be written inside notes about someone else — a summary of a conversation, or a reminder the user set. A member of our team reviews each of these by hand. Anything that is about you is deleted in full. We never edit or partly rewrite a stored fact: we remove it whole or leave it whole.

Conversation summaries. Where a summary described you, we generate a new summary from the messages that remain. If a fact about you still appears in it, we delete that summary entirely.

What we keep. The user's client record, and notes that mention you only in passing — for example, a reminder to send you a document — normally remain in their account. A passing mention can still be personal data, so we review each one against the user's legal basis for holding it and delete it where no exception applies.

What we cannot fully remove. Our search system turns each conversation into a numerical representation. Where a conversation included other people as well as you, that representation is derived from everyone's messages together and cannot be separated, so it is not deleted. It holds no readable text. We test whether it could reasonably be reconnected to an individual and record the result; if it can, we treat it as personal data and handle it accordingly.

How long it takes. We confirm receipt promptly and aim to complete an erasure within 30 days of confirming it. Part of the process is reviewed by a person, so if a request needs longer we will tell you why and when to expect it.

Proof that it happened. We keep a record that your request was handled — dates and counts only, with no name and none of the erased content — even if the user later closes their account.

11. Who else touches your data

We never share data between Tovira users and we never sell it. These suppliers help us run the service:

This section is kept current. We will update it, and email you, before adding any supplier that processes conversation content.

We may disclose data where the law requires it, or to establish or defend a legal claim.

12. International transfers

Your data is stored in Sweden (section 6). Anthropic and Groq process conversation content in the United States.

We only work with suppliers who are themselves subject to comprehensive data protection obligations in their own jurisdiction, and we require them by contract to protect your data to a standard consistent with this policy.

13. Your rights

You can ask us to give you a copy of your data, correct it, delete it, restrict or stop certain processing, object to processing based on legitimate interests, or provide it in a portable form.

You do not need to ask us for a copy. Export is built into Tovira and works at any time — including when your subscription has lapsed, when you have hit a usage limit, and while your account is being closed. We do not hold your data hostage.

For anything else, write to hello@tovira.io. We respond within 30 days.

If you are unhappy with our response you may complain to the UAE Data Office, the federal authority responsible for personal data protection in the UAE.

14. Security

Every account's data is isolated at the database level, not only in application code. Passwords are hashed. Traffic is encrypted in transit. Access to production systems is restricted and logged.

No system is perfectly secure. If a breach occurs that is likely to put your data at risk, we will notify you and the relevant regulator without undue delay, and in any event within any timeframe the law requires.

15. Children

Tovira is for professional use and is not intended for anyone under 18.

16. Changes

We will post changes here. Where a change is significant we will email you before it takes effect.